Abstract

It is well known that the alteration of one or more variables of the urinary steroid profile in comparison to a population-based or individual range may be caused by the administration of testosterone or its precursors. One way to prove the exogenous administration of testosterone or its precursors is the use of isotope ratio mass spectrometry (IRMS). In case IRMS remains inconclusive longitudinal analyses can be performed. At present, just univariate methods have been used to evaluate a potential ‘adverse analytical f1nding’ from the rest of the data:subject-based reference limits, outliers test, predictive model (Bayesian test), etc. Here we propose a multivariate analysis (MVA) approach for the longitudinal studies, taking into account the steroid profile and some metabolites ratios, analyzed globally. The goal is to build up a predictive model to judge whether a sample showed an abnormal profile using principal components analysis (PCA) as MVA method. Some authors have already proposed this kind of approach in the human antidoping area, but using population studies. In a first step we have analyzed samples collected from subjects of two different ethnic origins (Caucasians and Asians) treated with intramuscular testosterone enanthate to prove the effectiveness of the approach. In a second step, we applied the method to anonymous real longitudinal studies and the results were compared with the current WADA (World Antidoping Agency) approach. Einleitung